none
When will Microsoft release the facial recognition for Kinect for Windows? RRS feed

  • General discussion

  • When will Microsoft release the facial recognition for Kinect for Windows V2?

    You have released the facial recognition for Xbox ONE because it is essential for game developers BUT don't you think that Windows game developers using kinect for Windows v2 need a facial recognition api?

    It is ridiculous that kinect for xbox one has facial recognition but kinect for windows doesn't!Who made the decision of leaving facial recognition out of the Kinect for Windows? Very bad decision.

    I know at least 3 corporations that would have bought a few hundred thousand kinect for Windows if features like this were native. Imagine the revenue lost already by Microsoft because many companies are looking already at software that uses normal web cameras instead, for face and body tracking, emotion tracking and also facial recognition.

    No wonder Satia is sacking a lot of incompetents and old school guys that for whatever reason never think things through. I would have accepted if they said 'we didn't have time but will release it soon' BUT it has passed 5 months already and nothing was done.


    Saturday, December 13, 2014 9:02 AM

All replies

  • What is the scenario you are looking to support? There is no facial recognition Kinect api's on Xbox One. As a platform, the platform does have identity functionality but mainly for entertainment purposes. Developers don't get direct access to that.

    While it can recognize a small subset of people, like you and your immediate family members, if you are trying to scale that out to an office or building, that is not something that it was designed for. Also consider the scenario if it gets it wrong. On the Xbox, it asks if this is you and you can correct it. I don't think this would be acceptable for a security scenario.


    Carmine Sirignano - MSFT

    Monday, December 15, 2014 7:26 PM
  • Thanks for the reply.

    What made you think that it is about security or out of an office or building? I didn't mentioned security whatsoever!

    My post was very focus on the mistake made by Microsoft!

    The XBox One Kinect facial recognition of some kind (you can call it identity management or whatever) was not created to support only avatars/games BUT as you and most people know it was also to allow the control of specific funtions when it recognises someone e.g. Turn on the TV and tune into your preferred channel. So the API is quite extensive in terms of its functionality.

    With Kinect for Windows, did Microsoft think that nobody would use it to create interactive games for Windows for example? Or that nobody would ever need that sort of facial recognition API? You might say it is a limited one, I accept that, but everyone understands that Kinect was built for a controlled environment i.e. a living room so I really don't get why did someone at Microsoft decided that the "Avatar/Identity Management" API wasn't required by anyone who wanted the Kinect for Windows! WHAT??

    If I was one the VPs @ Microsoft, I would be asking a lot of questions...

    I think you guys need to make it a priority for the next SDK release and put the usual caviats - it is for a controlled environment where there are 5-6 people present, etc... Everyone will accept that! Make it extensible for other folks who want to do further development. Simples!


    Tuesday, December 16, 2014 12:46 PM
  • Nothing is stopping you from developing your own super duper awesome recognition system on top of the face, body and video data. Heck there may even be a market for it and an opportunity for you to get rich! :)

    In my opinion the SDK should focus on delivering the best raw sensor data and tracking data as possible.

    Subject recognition specifications can vary greatly depending on the user scenario (accuracy, size of target group, which data it uses etc). So it makes more sense to leave that up to the users instead of creating one generic recognizer that suits only a subset of user cases.


    Just my 2 cents....

    Brekel

    Tuesday, December 16, 2014 1:03 PM
    Moderator
  • So why was the XBox One Kinect developed with all that functionality?

    They could have left up to you and me to develop it, right?

    Your 2 cents don't make sense! :-) If the functionality didn't exist then I would agree with you and take your 2 cents! ;-)

    My point is simple, if that functionality already existed in the XBox One, then why was everything else ported apart from that one? Crazy decisions that never take the customer expectations and needs in consideration!

    Simple comparison - if you produce a car that has all the bells and whistles and then decide to produce a variation of that model but a car with an electric engine and then take out the integrated GPS from the new model, do you think you will meet customers expectations and needs? And that they will be happy? I think we know the answer!

    And of course, I could go to a store and buy a GPS system and stick it to my windscreen!!! But guess what? I wouldn't do that and I wouln't buy that model. Some people will! But now, you have limited your market!


    • Edited by Giorgio2007 Tuesday, December 16, 2014 1:17 PM
    Tuesday, December 16, 2014 1:16 PM
  • So Kinect v2 is the Kinect v1 but with a better hardware. Just this. If you want do something cool you always have to use opencv, openni ecc... They told us so much cool things you can do with kinect (face rec, heart monitor...) but the truth is that it wasn't made for be new a developer's device but just something you have to buy for play xbox.
    Best features has no api.
    My 2 cent is "worst microsoft business decision ever"
    • Edited by gioolio Wednesday, January 28, 2015 10:15 AM
    Wednesday, January 28, 2015 10:15 AM
  • I can see your point about wishing for parity between Xbox and  Windows versions, but I tend to agree with Brekel.

    It would not be hard to replicate the recognition routines, as you get the HD face and face which gives you feature points then just compare this vector of points with your library of users and find the nearest match. Then sell this to the 3 corporations you mentioned and make lots of money.

    As for the heart rate monitor, the code is out there, as I believe it originally came out of a paper from MIT using a standard webcam.

    If I want to go out and make a heart rate monitor or facial recognition application with the Kinect for use in a clinical setting for example, I would have to prove that it works and any problems that may arise from using it clinically would be my fault (and potentially be at odds with the 'no high risk' settings in the EULA). To me it makes sense why these easily DIY'able features haven't been released onto Windows. 

    Wednesday, January 28, 2015 11:59 AM
  • So what do you agree with Brekel exactly?

    That anyone with time and resources can do it? Well, tell me something I don't know!!! ANYTHING is possible when you have time, money and of course resources! I could develop a new device similar to Kinect! DO I have the time and the money? NO!

    You both are missing the point! Read my posts again.

    This is simply a very bad business decision made by my guess is the usual suspects! ;-)

    The problem seems to me that the XBox division is completelly autonomous and separata from Microsoft Windows guys! The integration never happened so this type of silly decisions were bound to happen!

    I would still sack some of the guys or would make them work their ass off and provide what is already Microsoft available libraries, read again, already available libraries and port that to Windows to be accessible through API calls!

    And wrap it up with usual T&Cs, use it as your own risk, blah blah... solicitors type of blurb... ;-)


    • Edited by Giorgio2007 Wednesday, January 28, 2015 2:55 PM
    Wednesday, January 28, 2015 2:54 PM
  • I agree that the SDK should focus on delivering the best raw data and tracking data as possible. With this anyone can implement the latest and most cutting edge computer vision algorithms available. Why would I want a facial recognition routine that was ok in 2014 when I could implement my own bleeding edge one in 2015? As smaller developers we have much more freedom to move and get things done than MS does. This is an advantage to us.
    Wednesday, January 28, 2015 3:18 PM
  • Why would Microsoft creat Windows? Or Word? Or Excel?

    You are also missing the point!

    But answering to your post, why would I want to code something from scratch when Microsoft already has developed a very good solution on the XBox? And creating an API that can be extended would allow you to enhance it and have your bleeding edge in 2015...whatever that means to you!

    Would you prefer Microsoft to develop Excel without the pivot functions? And then someone came along and say "it is right what they have done, because you can now develop your own pivot functions and have always a bleeding edge excel"!!

    I am not asking Microsoft to develop something new! They already have it on a commercial product!So why leave it out? It just doesn't make sense.

    Thursday, January 29, 2015 12:08 AM
  • Chrome and firefox extensions are good examples of how products can be enhanced and kept up to date by smaller teams of developers. Would I prefer my office to have more user generated "extensions"? Sure sounds fun, some may even do things better than the stock product.

    Back to topic, we disagree on what makes business sense, which is fine. The face recognition on the xbox is not that great. In my own experience it does not work all the time and refuses to recognise me if I grow a beard or get my hair cut. And that is just when it has to guess between 2 people on my user list. 

    Personal preferences and anecdotal evidence aside, who knows why it is not being released. Get an invite/buy a xbox dev license and try it out. Or wait for Windows 10 and hope that the promise of universal apps will open this up. Or write one/commission someone to make one for you if you can make a business case for it. 

    Thursday, January 29, 2015 8:46 AM
  • Any face regonition will not work if you grow a bear or start to use glasses, etc... if you have a biometric passport, try to grow a beard and then try to use any of the automate board controls kiosks around Europe! It just doesn't work! And believe me, they are highly advanced systems.

    Anyway, I have an XBox one since April 2014, got it free at the MS Conference in SF and I have tried it. Too expensive to have XBox One's as a business solution and really overkill! But my point was also, what about windows games? Kinect for windows was not thought through, full stop!

    Friday, January 30, 2015 12:32 AM
  • Ok, I understand your point guys and if my kinect detection/recognition software is for a "Bank access" or some medical surgery, you are 100% right.
    BUT maybe i don't need such precision, maybe i just need a simple smooth joint API because i have to move apple.png on buket.png in my app and i can't waste my time trying to understand how to create a joint move prediction.
    I will if i have to launch a Nasa shuttle in the space.

    So I can replicate the recognition routines if it's worth it, but maybe i don't need the 100% accuracy.

    For the question "Raw is better", would be better buy an hd camera and IR trasmitter and reciver and build your personal kinect? Or is usefull that someone already did it for you? is the same for the recognition feature. No one said we want a "DNA test" accuracy, but would be nice release the xbox recognition feature as api (even without the open source code).
    Because i payed the Microsoft Kinect and i payed the Microsoft Windows 8.1, i think i have acquired the right to use something more then raw things.

    (I.g. My software use kinect to move the mouse cursor, but i can't use the kinect region just because i'm not using an win xaml app. Why? You made that api for developers. So give it to me the "raw" kinect region! No. I have to code my personal kinect region and phiz.)

    Do you think that after all this coding suffer i will create a win app store or i will try to be the most platform indipendent possible?
    Microfost's is such a bad business decision; if you give me tools and raw data will be easier and faster develope software for you. But if i have to do all the job, you can forget the win app store development, why i have to give you this vantage for free?
    I want to code for the win app store but i have no tools/api. You want to make your medical/recognition system and this will never be on win app store, Microsoft should help me because i create more value for them then you.
    "You can tell me what to do or how to do it, but not both"
    • Edited by gioolio Wednesday, February 4, 2015 11:31 AM
    Wednesday, February 4, 2015 11:28 AM
  • We seem to be going slightly off topic towards 'why aren't Microsoft...' rather than 'when will Microsoft...' and since neither of us are informed of internal Microsoft decisions I don't know if anything more can be added to this discussion. 
    Wednesday, February 4, 2015 5:45 PM
  • Yes why would someone want a working API written and supported by a billion dollar company when I can be forced to manage my own and update it, instead of thinking about how to write the app I want.

    Also look at Project Oxford now. 

    Mark Rowe MCPD:Windows, MCPD:Web, MCPD:Azure, MCTS:Biztalk http://MicrosoftIntegrationArchitect.com

    Friday, November 20, 2015 12:34 AM